bandit

Human-friendly unit testing for C++11

Running tests

With bandit, you create a command line application that runs all registered tests, reports the result on stdout, and then returns with exit code 0 if all tests passed, or exit code greater than 0 if some tests did not pass.

Command line arguments

Bandit enables you to tweak the way tests are performed and reported by specifying command line options. Running [name of your executable] --help will list the available options.

$ specs --help
USAGE: <executable> [options]

Options:
  --version,               Print version of bandit
  --help,                  Print usage and exit.
  --reporter=<reporter>,   Select reporter: crash, dots, singleline, xunit,
                           info, spec
  --no-color,              Suppress colors in output
  --formatter=<formatter>, Select error formatter: default, vs
  --skip=<substring>,      Skip all 'describe' and 'it' containing substring
  --only=<substring>,      Run only 'describe' and 'it' containing substring
  --break-on-failure,      Stop test run on first failing test
  --dry-run,               Skip all tests. Use to list available tests

Running a subset of the tests

You can skip tests from within your source code by using the describe_skip(), xdescribe(), it_skip(), and xit() constructs, or by setting the last (optional) argument of it() and describe() to true.

You can skip tests from the command line by using the --skip=<substring> option. By using this, you tell bandit to skip running all describe() and it() whose names contain <substring>.

It is also possible to tell bandit to only run a subset of the tests by passing --only=<substring>. This will cause bandit to only run those describe() and it() whose names contain <substring>.

You can use --only and --skip in combination and multiple times to filter the running tests according to your needs.

Tweaking the output

Reporters

You can specify the way bandit reports the progress and results of a test by using the --reporter option:

--reporter=dots

Reports each executed test as a dot (.) or as an ā€˜Fā€™ or ā€˜Eā€™ depending on whether it succeeds or not.

This is the default.

--reporter=singleline

Reports the progress on a single line. Updating counters of executed, failed and skipped tests.

Very useful if you do not want to be bothered a lot by output of the testing framework.

--reporter=xunit

Reports the test results in a XML format supported by most continuous integration servers.

--reporter=spec

Reports the tests in a format similar to how they are written along with the status of each test.

--reporter=info

Reports the tests in an appealing form with a marker for every relevant it. The marker says [ TEST ] when the test starts, [ PASS ] when the test passed and [ FAIL ] when the test failed. describe()s are declared by begin and end and every end contains a summary. Skipped describe()s are not shown at all. This reporter is somewhat inspired by the Google Test framework.

--reporter=crash

Reports one it() per line containing the whole describe() stack. Very useful, for example, if you have a huge amount of tests and you are testing code that may crash, in a continuous integration system that sends you a mail with the last few lines of output.

Formatters

With the --formatter=<formatter> option you can specify how bandit reports failed assertions. Different IDEs expect errors to be reported in different ways to be able to quickly navigate to the location of a failed assertion.

--formatter=default

The default formatter reports errors as

<filename>:<line>: <error message>

This matches how gcc and clang reports errors. As the name indicates, this formatter is the default.

--formatter=vs

The Visual Studio formatter reports errors the way Visual Studio wants errors to be reported, as

<filename>(<line>): <error message>

Colors

By default, bandit uses colors to highlight the status of the current test run. In certain IDEs or when running tests in continuous integration environment, this may look ugly. By specifying --no-color you can tell bandit to stop using colors.

Tweaking the behavior

Exit on first error

The option --break-on-failure exits bandit when the first error or failed test occurs.

Dry run

The --dry-run option skips every test. In combination with --reporter=spec, this can be used to get a list of the available tests.